Structure and Performance of Decision Support Algorithms on ActiveDisks

Growth and usage trends for large decision support databases indicate thatthere is a need for architectures that scale the processing power as thedataset grows. These trends indicate that the processing demand for largedecision support databases is growing faster than the improvement inperformance of commodity processors. To meet this need, several researchershave recently proposed Active Disk/IDISK architectures which integratesubstantial processing power and memory into disk units. In this paper, weexamine the utility of Active Disks for decision support databases. We try toanswer the following questions. First, is it possible to restructurealgorithms for common decision support tasks to utilize Active Disks? Second,how does the performance of Active Disks compare with that of traditionalservers for these tasks? Finally, how would Active Disks be integrated intothe software architecture of decision support databases?